GPT-4o-compliant relational AI models are now accessible to AWS subscribers. AWS announced six new unit sizes for various use situations in the fresh Amazon Nova home on December 3 at the AWS are: Invent event held in Las Vegas and online.
” Inside Amazon, we have about 1, 000 generative AI applications in motion, and we’ve had a bird ‘s-eye view of what application builders are still grappling with”, Rohit Prasad, SVP of Amazon Artificial General Intelligence, said in the press release.
” Our new Amazon Nova models are intended to help with these challenges for internal and external builders and provide compelling intelligence and content generation while also delivering meaningful progress on latency, cost-effectiveness, customization, Retrieval Augmented Generation (RAG ), and agentic capabilities”.
What is Amazon Nova?
AWS’s Amazon Bedrock AI having service includes a range of relational AI base models. Companies can now test with three length options:
- A fast response period of 210 production tokens per second makes the Amazon Nova Micro a text-only design. Amazon claims it outperforms Meta’s Llama 3.1 8B and Google’s Gemini 1.5 Flash-8B. Applications that demand swift responses for relatively little money are the candidates for Nova Micro.
- Another member of the Nova home, the Amazon Nova Lite. Unlike Micro, it does analyze either photo, picture, or word inputs. Similar to OpenAI’s GPT-4o small, Nova Lite is intended for rapid summarization and understanding of graphs or video lectures. Amazon Nova Lite is suitable for some quasi-autonomous chained actions used for” AI agent” things because it can interpret photographs on computer monitors and perform work calling.
- Amazon Nova Pro is the mid-range design. Amazon said it performs faster, more effectively, and costs less than OpenAI’s GPT-4o or Google’s Gemini 1.5 Pro. Nova Pro supports agentic procedures and can interpret words or images.
When customers have a Nova design, they may fine-tune it based on their amazing data.
Organizations can choose from an image generation model ( Amazon Nova Canvas ) and a video model ( Amazon Nova Reels ) in addition to the size options. Both of these are intended to create” studio-quality” content.
- Based on text or graphic causes, Nova Canvas creates photos. Amazon points out that the product has security features like willing guardrails and watermarking.
- Amazon intends to increase the potential video length to two hours in” the approaching months,” according to Nova Reels, who creates six-second videos.
Notice: AI regulation is ongoing in Australia, with a commission calling for big models from OpenAI, Meta, and Google to qualify as “high-risk”.
What’s future?
The third type in the Nova range, Nova Premier, will not be accessible until the first third of 2025. Amazon expects Nova Premier to bring multimodal (video, image, or text-to-text ) interpretation and a hefty data collection that businesses can use to train other types.
Moreover, Amazon plans to add a design that can respond normally to spoken discussion. They are also working on a multimodal-to-multimodal concept to perceive and production text, pictures, videos, or music.
While it’s yet too early to see how Nova will compete with rivals like OpenAI, Google, and Meta, Amazon scored one major partner in SAP, which offers the models on its AI Core platform.
TechRepublic is covering AWS re: Invent remotely.
